The use of canonical analysis

The first part of this non-technical review of canonical analysis is concerned with the principle, the data requirements, the interpretation, the evaluation and the application in consumer research of canonical analysis. In this context, concepts as canonical correlation, weights, loadings and scores are explained. Their interrelationships are discussed. Then the characteristics of specific forms of canonical analysis: canonical correlation, canonical regression, redundancy analysis and partial canonical analysis are discussed. Their different application properties are emphasized. The last part consists of an application of canonical correlation analysis for brand positioning. Here the relationship with discriminant analysis is illustrated. Furthermore the usage of canonical analysis for optimal scaling purposes is illustrated for the same example.